In their insightful article, Kenneth Kendler, Peter Zachar and Carl Craver recommend the programmatic modelling of psychiatric disorders as kinds of mechanistic property clusters (MPC) (Kendler et al. Reference Kendler, Zachar and Craver2010). According to this view, mental disorders are individuated by the whole cluster of mechanisms involved in the causation of their respective clinical syndromes. As the authors assert, ‘the identity of the disease […] is grounded in the similarity of the complex, mutually reinforcing network of causal mechanisms in each case’ (p. 6). However, on the same page a few lines below, they also claim that since ‘the same cluster of symptoms might arise from different mechanisms’, ‘MPC kinds are […] “multiply realizable” ‘. I find this claim inconsistent with their previously cited assertion. If the identity of MPC kinds is grounded in the complex network of their causative mechanisms, then they cannot be ‘multiply realizable’. What are ‘multiply realizable’ are not MPC kinds, but clusters of strongly similar clinical signs/symptoms. Instead, the MPC model clearly implies, however, that mental disorders sharing the same cluster of clinical signs/symptoms, emerging through different networks of causative mechanisms, constitute distinct MPC kinds, each one uniquely ‘realized’. Besides, this view is already – although only implicitly and rudimentarily – endorsed in contemporary psychiatric classification systems. More precisely, the endorsement of this view is reflected in their systematic distinction of presumably ‘primary’ mental disorders of still unknown causation from ‘secondary’ mental disorders due to diagnosable general medical conditions, irrespective of the otherwise possibly strong similarities in their clinical manifestations.
Declaration of Interest
None.
The authors reply
Our essay addresses two closely related but distinct questions. The first question is ‘What does it mean to say of a given psychiatric disorder that it is a kind of disorder rather than, for example, a more or less arbitrary collection of properties that happen to co-occur?’ To this, we answer that kinds of disorders, or the things we ought to count as kinds, are not arbitrary collections of individuals but rather property clusters in which the properties tend to co-occur because there exists some underlying mechanism that explains their co-occurrence. Nothing in this view of kinds demands that there should be only one underlying mechanism, either in a single afflicted individual or across a population. Kinds are regular clusters of properties sustained by one or more mechanisms such that they are regular, repeatable, predictable, and at least potentially manipulable. The second question is: What individuates kinds of psychiatric disorders? What makes one kind different from another? How do we know when we have one kind as opposed to two or more hidden under the same label? In answer to that, it is tempting to say (beyond the answer to the first question) that kinds correspond in a one-to-one way with mechanisms: if there are two kinds of mechanism for one property cluster, then there are in fact two higher-level kinds, one for each mechanism. This answer to the second question, however, would be inconsistent with multiple realizability, as multiple realizability is just the idea that the same kind can be realized by different mechanisms. Our essay defends the MPC as an answer to the first question while trying to resist the claim that it provides a complete answer to the second. Given that differences in underlying mechanisms often do signal differences in property clusters, it makes sense to sort psychiatric disorders in part on the basis of what we learn about their biological mechanisms. Superficially similar syndromes might be underwritten by different underlying mechanisms, but slightly different mechanisms could be seen as producing highly similar outcomes. However, it was not a goal of our essay to deal more deeply with this thorny and long-debated issue – that is the relative merits and practicalities of using ‘surface features (i.e. symptoms and signs) or etiology (=mechanism) to undergird a psychiatric nosology.
